A Deep Neural Network for Estimating Low-Density Lipoprotein Cholesterol From Electronic Health Records: Real-Time Routine Clinical Application.
Ontology highlight
ABSTRACT: Background
Previously, we constructed a deep neural network (DNN) model to estimate low-density lipoprotein cholesterol (LDL-C).Objective
To routinely provide estimated LDL-C levels, we applied the aforementioned DNN model to an electronic health record (EHR) system in real time (deep LDL-EHR).Methods
The Korea National Health and Nutrition Examination Survey and the Wonju Severance Christian Hospital (WSCH) datasets were used as training and testing datasets, respectively. We measured our proposed model's performance by using 5 indices, including bias, root mean-square error, P10-P30, concordance, and correlation coefficient. For transfer learning (TL), we pretrained the DNN model using a training dataset and fine-tuned it using 30% of the testing dataset.Results
Based on 5 accuracy criteria, deep LDL-EHR generated inaccurate results compared with other methods for LDL-C estimation. By comparing the training and testing datasets, we found an overfitting problem. We then revised the DNN model using the TL algorithms and randomly selected subdata from the WSCH dataset. Therefore, the revised model (DNN+TL) exhibited the best performance among all methods.Conclusions
Our DNN+TL is expected to be suitable for routine real-time clinical application for LDL-C estimation in a clinical laboratory.
SUBMITTER: Hwang S
PROVIDER: S-EPMC8371492 | biostudies-literature |
REPOSITORIES: biostudies-literature
Similar Datasets
Project description:BACKGROUND:Low-density lipoprotein cholesterol (LDL-C) response to statin therapy has not been fully elucidated in real-world populations. The primary objective of this study was to characterize statin LDL-C dose-response and its heritability in a large, multiethnic population of statin users. METHODS:We determined the effect of statin dosing on lipid measures utilizing electronic health records in 33?139 statin users from the Kaiser Permanente GERA cohort (Genetic Epidemiology Research on Adult Health and Aging). The relationship between statin defined daily dose and lipid parameter response (percent change) was determined. RESULTS:Defined daily dose and LDL-C response was associated in a log-linear relationship (?, -6.17; SE, 0.09; P<10-300) which remained significant after adjusting for prespecified covariates (adjusted ?, -5.59; SE, 0.12; P<10-300). Statin type, sex, age, smoking status, diabetes mellitus, and East Asian race/ethnicity were significant independent predictors of statin-induced changes in LDL-C. Based on a variance-component method within the subset of statin users who had at least 1 first-degree relative who was also a statin user (n=1036), heritability of statin LDL-C response was estimated at 11.7% (SE, 8.6%; P=0.087). CONCLUSIONS:Using electronic health record data, we observed a statin LDL-C dose-response consistent with the rule of 6% from prior clinical trial data. Clinical and demographic predictors of statin LDL-C response exhibited highly significant but modest effects. Finally, statin-induced changes in LDL-C were not found to be strongly inherited. Ultimately, these findings demonstrate (1) the utility of electronic health records as a reliable source to generate robust phenotypes for pharmacogenomic research and (2) the potential role of statin precision medicine in lipid management.
| S-EPMC6214660 | biostudies-literature
Project description:BackgroundRoutinely recorded electronic health records (EHRs) from general practitioners (GPs) are increasingly available and provide valuable data for estimating incidence and prevalence rates of diseases in the population. This paper describes how we developed an algorithm to construct episodes of illness based on EHR data to calculate morbidity rates.ObjectiveThe goal of the research was to develop a simple and uniform algorithm to construct episodes of illness based on electronic health record data and develop a method to calculate morbidity rates based on these episodes of illness.MethodsThe algorithm was developed in discussion rounds with two expert groups and tested with data from the Netherlands Institute for Health Services Research Primary Care Database, which consisted of a representative sample of 219 general practices covering a total population of 867,140 listed patients in 2012.ResultsAll 685 symptoms and diseases in the International Classification of Primary Care version 1 were categorized as acute symptoms and diseases, long-lasting reversible diseases, or chronic diseases. For the nonchronic diseases, a contact-free interval (the period in which it is likely that a patient will visit the GP again if a medical complaint persists) was defined. The constructed episode of illness starts with the date of diagnosis and ends at the time of the last encounter plus half of the duration of the contact-free interval. Chronic diseases were considered irreversible and for these diseases no contact-free interval was needed.ConclusionsAn algorithm was developed to construct episodes of illness based on routinely recorded EHR data to estimate morbidity rates. The algorithm constitutes a simple and uniform way of using EHR data and can easily be applied in other registries.
| S-EPMC6688441 | biostudies-literature
Project description:BackgroundLow-density lipoprotein cholesterol (LDL-C) is a risk factor for atherosclerotic cardiovascular disease (ASCVD). There are limited real-world data on LDL-C lowering with evolocumab in United States clinical practice.HypothesisWe assessed LDL-C lowering during 1 year of evolocumab therapy.MethodsThis retrospective cohort study used linked laboratory (Prognos) and medical claims (IQVIA Dx/LRx and PharMetrics Plus® ) data. Patients with a first fill for evolocumab between 7/1/2015 and 10/31/2019 (index event) and LDL-C ≥ 70 mg/dL were included (overall cohort; N = 5897). Additionally, a patient subgroup with a recent myocardial infarction (MI) within 12 months (median 130 days) before the first evolocumab fill was identified (N = 152). Reduction from baseline LDL-C was calculated based on the lowest LDL-C value recorded during a 12-month follow-up period.ResultsThe mean (SD) age was 65 (10) years; 61.9% of patients had ASCVD diagnoses and 70.7% of patients were in receipt of lipid-lowering therapy. Following evolocumab treatment, changes in LDL-C from baseline were -60% in the overall cohort (median [interquartile range (IQR)] 146 [115-180] mg/dL to 58 [36-84] mg/dL) and -65% in the recent MI subgroup (median [IQR] 137 [109-165] mg/dL to 48 [30-78] mg/dL). In the overall cohort and recent MI subgroup, 62.1% and 69.7% of patients achieved LDL-C < 70 mg/dL, respectively.ConclusionsIn this real-world analysis, evolocumab was associated with significant reductions in LDL-C comparable to that seen in the FOURIER clinical trial, which were durable over 1 year of treatment.
| S-EPMC8119857 | biostudies-literature
Project description:BackgroundLimited data are available for validation of low-density lipoprotein cholesterol (LDL) calculation (LDLcal) in the adult Korean population. The aim of this study was to develop and validate a new equation for LDLcal and to compare it with previous such equations in a Korean population.MethodsA new equation for LDLcal was developed (LDLChoi). LDLChoi and 11 other previously published equations were applied and compared with directly measured LDL concentration (LDLdirect) in a development cohort (population 1), an independent validation cohort in the same laboratory (population 2), and the Korea National Health and Nutrition Examination Survey 2017 cohort (population 3).ResultsAmong the 12 equations, the newly-developed equation (LDLChoi = total cholesterol - 0.87 x high-density lipoprotein cholesterol - 0.13 x triglycerides) had the highest intraclass correlation coefficient (ICC) and the lowest mean systemic difference and median absolute percentage error in populations 1 and 2 but not in population 3. Subgroup analysis showed good agreement between LDLChoi and LDLdirect (ICC > 0.75) in population 2, whose LDLdirect < 70 mg/dL. For samples with high triglycerides (> 400 mg/dL), equation accuracy varied. Categorization concordance according to the National Cholesterol Education Program Adult Treatment Panel III criteria with the other 11 equations were less than 80%; that of LDLChoi was 87.6 and 87.4% in populations 1 and 2, respectively.ConclusionsAccuracy of 12 equations for LDLcal varied by cohort and subgroup based on LDLdirect and triglycerides. A laboratory-specific equation for LDLcal and/or LDLdirect may be needed for accurate evaluation of LDL status.
| S-EPMC8453999 | biostudies-literature
Project description:BackgroundDirect feedback on quality of care is one of the key features of a learning health care system (LHS), enabling health care professionals to improve upon the routine clinical care of their patients during practice.ObjectiveThis study aimed to evaluate the potential of routine care data extracted from electronic health records (EHRs) in order to obtain reliable information on low-density lipoprotein cholesterol (LDL-c) management in cardiovascular disease (CVD) patients referred to a tertiary care center.MethodsWe extracted all LDL-c measurements from the EHRs of patients with a history of CVD referred to the University Medical Center Utrecht. We assessed LDL-c target attainment at the time of referral and per year. In patients with multiple measurements, we analyzed LDL-c trajectories, truncated at 6 follow-up measurements. Lastly, we performed a logistic regression analysis to investigate factors associated with improvement of LDL-c at the next measurement.ResultsBetween February 2003 and December 2017, 250,749 LDL-c measurements were taken from 95,795 patients, of whom 23,932 had a history of CVD. At the time of referral, 51% of patients had not reached their LDL-c target. A large proportion of patients (55%) had no follow-up LDL-c measurements. Most of the patients with repeated measurements showed no change in LDL-c levels over time: the transition probability to remain in the same category was up to 0.84. Sequence clustering analysis showed more women (odds ratio 1.18, 95% CI 1.07-1.10) in the cluster with both most measurements off target and the most LDL-c measurements furthest from the target. Timing of drug prescription was difficult to determine from our data, limiting the interpretation of results regarding medication management.ConclusionsRoutine care data can be used to provide feedback on quality of care, such as LDL-c target attainment. These routine care data show high off-target prevalence and little change in LDL-c over time. Registrations of diagnosis; follow-up trajectory, including primary and secondary care; and medication use need to be improved in order to enhance usability of the EHR system for adequate feedback.
| S-EPMC7163416 | biostudies-literature
Project description:Background Accurate measurement of the cholesterol within lipoprotein(a) (Lp[a]-C) and its contribution to low-density lipoprotein cholesterol (LDL-C) has important implications for risk assessment, diagnosis, and treatment of atherosclerotic cardiovascular disease, as well as in familial hypercholesterolemia. A method for estimating Lp(a)-C from particle number using fixed conversion factors has been proposed (Lp[a]-C from particle number divided by 2.4 for Lp(a) mass, multiplied by 30% for Lp[a]-C). The accuracy of this method, which theoretically can isolate "Lp(a)-free LDL-C," has not been validated. Methods and Results In 177 875 patients from the VLDbL (Very Large Database of Lipids), we compared estimated Lp(a)-C and Lp(a)-free LDL-C with measured values and quantified absolute and percent error. We compared findings with an analogous data set from the Mayo Clinic Laboratory. Error in estimated Lp(a)-C and Lp(a)-free LDL-C increased with higher Lp(a)-C values. Median error for estimated Lp(a)-C <10 mg/dL was -1.9 mg/dL (interquartile range, -4.0 to 0.2); this error increased linearly, overestimating by +30.8 mg/dL (interquartile range, 26.1-36.5) for estimated Lp(a)-C ≥50 mg/dL. This error relationship persisted after stratification by overall high-density lipoprotein cholesterol and high-density lipoprotein cholesterol subtypes. Similar findings were observed in the Mayo cohort. Absolute error for Lp(a)-free LDL-C was +2.4 (interquartile range, -0.6 to 5.3) for Lp(a)-C<10 mg/dL and -31.8 (interquartile range, -37.8 to -26.5) mg/dL for Lp(a)-C≥50 mg/dL. Conclusions Lp(a)-C estimations using fixed conversion factors overestimated Lp(a)-C and subsequently underestimated Lp(a)-free LDL-C, especially at clinically relevant Lp(a) values. Application of inaccurate Lp(a)-C estimations to correct LDL-C may lead to undertreatment of high-risk patients.
| S-EPMC9238537 | biostudies-literature
Project description:Blood concentrations of lipoproteins and lipids are heritable risk factors for cardiovascular disease. Using genome-wide association data from three studies (n = 8,816 that included 2,758 individuals from the Diabetes Genetics Initiative specific to the current paper as well as 1,874 individuals from the FUSION study of type 2 diabetes and 4,184 individuals from the SardiNIA study of aging-associated variables reported in a companion paper in this issue) and targeted replication association analyses in up to 18,554 independent participants, we show that common SNPs at 18 loci are reproducibly associated with concentrations of low-density lipoprotein (LDL) cholesterol, high-density lipoprotein (HDL) cholesterol, and/or triglycerides. Six of these loci are new (P < 5 x 10(-8) for each new locus). Of the six newly identified chromosomal regions, two were associated with LDL cholesterol (1p13 near CELSR2, PSRC1 and SORT1 and 19p13 near CILP2 and PBX4), one with HDL cholesterol (1q42 in GALNT2) and five with triglycerides (7q11 near TBL2 and MLXIPL, 8q24 near TRIB1, 1q42 in GALNT2, 19p13 near CILP2 and PBX4 and 1p31 near ANGPTL3). At 1p13, the LDL-associated SNP was also strongly correlated with CELSR2, PSRC1, and SORT1 transcript levels in human liver, and a proxy for this SNP was recently shown to affect risk for coronary artery disease. Understanding the molecular, cellular and clinical consequences of the newly identified loci may inform therapy and clinical care.
| S-EPMC2682493 | biostudies-literature
Project description:Background MEDI6012 is recombinant human lecithin cholesterol acyltransferase, the rate-limiting enzyme in reverse cholesterol transport. Infusions of lecithin cholesterol acyltransferase have the potential to enhance reverse cholesterol transport and benefit patients with coronary heart disease. The purpose of this study was to test the safety, pharmacokinetic, and pharmacodynamic profile of MEDI6012. Methods and Results This phase 2a double-blind study randomized 48 subjects with stable coronary heart disease on a statin to a single dose of MEDI6012 or placebo (6:2) (NCT02601560) with ascending doses administered intravenously (24, 80, 240, and 800 mg) and subcutaneously (80 and 600 mg). MEDI6012 demonstrated rates of treatment-emergent adverse events that were similar to those of placebo. Dose-dependent increases in high-density lipoprotein cholesterol were observed with area under the concentration-time curves from 0 to 96 hours of 728, 1640, 3035, and 5318 should be: mg·h/mL in the intravenous dose groups and 422 and 2845 mg·h/mL in the subcutaneous dose groups. Peak mean high-density lipoprotein cholesterol percent change was 31.4%, 71.4%, 125%, and 177.8% in the intravenous dose groups and 18.3% and 111.2% in the subcutaneous dose groups, and was accompanied by increases in endogenous apoA1 (apolipoprotein A1) and non-ATP-binding cassette transporter A1 cholesterol efflux capacity. Decreases in apoB (apolipoprotein B) were observed across all dose levels and decreases in atherogenic small low-density lipoprotein particles by 41%, 88%, and 79% at the 80-, 240-, and 800-mg IV doses, respectively. Conclusions MEDI6012 demonstrated an acceptable safety profile and increased high-density lipoprotein cholesterol, endogenous apoA1, and non-ATP-binding cassette transporter A1 cholesterol efflux capacity while reducing the number of atherogenic low-density lipoprotein particles. These findings are supportive of enhanced reverse cholesterol transport and a functional high-density lipoprotein phenotype. Registration URL: https://www.clinicaltrials.gov; Unique identifier: NCT02601560.
| S-EPMC8403308 | biostudies-literature
Project description:Background Conventional "low-density lipoprotein cholesterol (LDL-C)" assays measure cholesterol content in both low-density lipoprotein and lipoprotein(a) particles. To clarify the consequences of this methodological limitation for clinical care, our study aimed to compare associations of "LDL-C" and corrected LDL-C with risk of cardiovascular disease and to assess the impact of this correction on the classification of patients into guideline-recommended LDL-C categories. Methods and Results Lipoprotein(a) cholesterol content was estimated as 30% of lipoprotein(a) mass and subtracted from "LDL-C" to obtain corrected LDL-C values (LDL-Ccorr30). Hazard ratios for cardiovascular disease (defined as coronary heart disease, stroke, or coronary revascularization) were quantified by individual-patient-data meta-analysis of 5 statin landmark trials from the Lipoprotein(a) Studies Collaboration (18 043 patients; 5390 events; 4.7 years median follow-up). When comparing top versus bottom quartiles, the multivariable-adjusted hazard ratio for cardiovascular disease was significant for "LDL-C" (1.17; 95% CI, 1.05-1.31; P=0.005) but not for LDL-Ccorr30 (1.07; 95% CI, 0.93-1.22; P=0.362). In a routine laboratory database involving 531 144 patients, reclassification of patients across guideline-recommended LDL-C categories when using LDL-Ccorr30 was assessed. In "LDL-C" categories of 70 to <100, 100 to <130, 130 to <190, and ?190 mg/dL, significant proportions (95% CI) of participants were reassigned to lower LDL-C categories when LDL-Ccorr30 was used: 30.2% (30.0%-30.4%), 35.1% (34.9%-35.4%), 32.9% (32.6%-33.1%), and 41.1% (40.0%-42.2%), respectively. Conclusions "LDL-C" was associated with incident cardiovascular disease only when lipoprotein(a) cholesterol content was included in its measurement. Refinement in techniques to accurately measure LDL-C, particularly in patients with elevated lipoprotein(a) levels, is warranted to assign risk to the responsible lipoproteins.
| S-EPMC7763787 | biostudies-literature
Project description:We report a study of machine learning applied to the phenotyping of psychiatric diagnosis for research recruitment in youth depression, conducted with 861 labelled electronic medical records (EMRs) documents. A model was built that could accurately identify individuals who were suitable candidates for a study on youth depression.Our objective was a model to identify individuals who meet inclusion criteria as well as unsuitable patients who would require exclusion.Our methods included applying a system that coded the EMR documents by removing personally identifying information, using two psychiatrists who labelled a set of EMR documents (from which the 861 came), using a brute force search and training a deep neural network for this task.According to a cross-validation evaluation, we describe a model that had a specificity of 97% and a sensitivity of 45% and a second model with a specificity of 53% and a sensitivity of 89%. We combined these two models into a third one (sensitivity 93.5%; specificity 68%; positive predictive value (precision) 77%) to generate a list of most suitable candidates in support of research recruitment.Our efforts are meant to demonstrate the potential for this type of approach for patient recruitment purposes but it should be noted that a larger sample size is required to build a truly reliable recommendation system.Future efforts will employ alternate neural network algorithms available and other machine learning methods.
| S-EPMC5566092 | biostudies-other